Anal. Chem. 2003, 75, 4686-4690
Hydrodynamic Micromanipulation of Individual Cells onto Patterned Attachment Sites on Biomicroelectromechanical System Chips Makoto Yoshida, Koji Tohda, and Miklo´s Gratzl*
Department of Biomedical Engineering, Case Western Reserve University, Cleveland, Ohio 44106
We report on a simple methodology to move selected single live cells to a desired location on a flat substrate, such as a patterned biomicroelectromechanical system chip. A macroscopic syringe-and-tube-based hydrodynamic manipulation system is used to achieve controlled cell navigation onto hydrophilic sites for cell attachment. Centimeter-per-second flow velocities generated by the system get downgraded to micrometers-per-second flow at the height of settled cells as a result of viscous flow in the medium. By pushing/pulling two syringes that produce two orthogonal flows, fine manipulation in any horizontal direction is feasible. After attachment of the desired cell(s) onto the selected hydrophilic site, all other unwanted cells are washed away from the surrounding hydrophobic surface with faster flow. This simple methodology is applicable for rapid cell pattern formation with high precision. Manipulation and positioning of individual cells are required to assist cell separation, cell microsurgery, gene transfection, drug injection, and many other procedures in cellular engineering. Availability of such techniques is also important to aid in constructing precisely defined cell structures in cellular transport studies. The need for this arose in our former studies that indicated that reproducible and quantitatively reliable data on cellular transport can only be obtained when both cell and microsensor patterns are precisely defined.1,2 The reason is that cell-cell interactions, as well as mass transport dynamics between cells and sensors, sensitively depend on the geometry of the cell preparation and sensor layout. Several types of cell manipulating methodology have been designed and successfully tested in different settings.3-8 Mechanical, electrical, and optical tools are used in the major approaches. Mechanical manipulations involve a micropipet or microtweezer. 3 Although this approach is simple and inexpensive, it does not provide for easy and gentle handling of single cells. Electrical (1) (2) (3) (4) (5) (6) (7) (8)
Yi, C.; Gratzl, M. Biophys. J. 1998, 75, 2255-2261. Lu, H.; Gratzl, M. Anal. Chem. 1999, 71, 2821-2830. Hochmuth, R. M. J. Biomech. 2000, 33 (1), 15-22. Schnelle, T.; Mu ¨ ller, T.; Hagedorn, R.; Voigt, A.; Fuhr, G. BBA-Gen Subjects 1999, 1428, 99-105. Ogata, S.; Yasukawa, T.; Matsue, T. Bioelectrochemistry 2001, 54, 33-37. Ashkin, A.; Dziedzic, J. M.; Yamane, T. Nature 1987, 330, 769-771. Ulanowski, Z.; Ludlow, I. K. Meas. Sci. Technol. 2000, 11, 1778-1785. Arai, F.; Ichikawa, A.; Ogawa, M.; Fukuda, T.; Horio, K.; Itoigawa, K. Electrophoresis 2001, 22, 283-288.
4686 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
manipulation utilizes dielectophoretic gradient forces.4,5 In inhomogeneous high-frequency electromagnetic fields, cells become polarized with respect to the surrounding aqueous medium, and therefore, they experience net dielectrophoretic effects. The resulting attractive or repulsive forces can be used for cell manipulation. Optical manipulation uses a laser beam whose electromagnetic gradient force allows the positioning of cells.6-8 Although both electrical and optical trapping can be used to manipulate single cells with good geometrical accuracy, the cost and complexity of the necessary equipment prevented their broader use so far. To realize easier and at the same time much less expensive cell manipulation, we introduce here a novel hydrodynamic cell micromanipulation methodology that can be operated manually using simple laboratory syringes and tubes. This technique is also gentler on the cells, as compared to other methodologies. To navigate selected cells onto desired sites under the microscope, accurate and precise movement control of individual suspended cells with a speed of a few micrometers per second is achieved by using simple hydrodynamic principles. In a related work, we designed Cell Interrogator chips9 that use relative hydrophilicity/hydrophobicity of surfaces patterned with microelectromechanical systems (MEMS) technologies,10 to facilitate cell attachment at the desired locations, such as sensing sites, as well as cell rejection on the rest of the chip. Feasibility of selective cell adhesion has been proved to work very well in experiments involving cell culture incubation times on the order of a few days.11 Recently, we demonstrated that a similar approach is also available when selective and site-specific cell adhesion is required to occur within several minutes,12 as on a Cell Interrogator chip. Adhesion onto surface-oxidized silicon (Si/SiO2) was found distinctively stronger than that onto a polyimide-coated substrate, even after only a few minutes of contact. It is in this context that we demonstrate the design and use of our syringeand-tube cell manipulation system. EXPERIMENTAL SECTION Apparatus. Cell Interrogator chips as used in this study consist of a series of micrometer-scale, thin film Pt ring electrodes (9) Yoshida, M.; Zorman, C.; Yang, J.; Tohda, K.; Gratzl, M. To be submitted for publication. (10) Lyshevski, S. E. MEMS and NEMS: systems, devices, and structures; CRC Press: Boca Raton, FL, 2002. (11) Hanein, Y.; Pan, Y. V.; Ratner, B. D.; Denton, D. D.; Bohringer, K. F. Sens. Actuators, B 2001, 81, 49-54. (12) Yoshida, M.; Zorman, C.; Yang, J.; Tohda, K.; Gratzl, M. Submitted. 10.1021/ac030055u CCC: $25.00
© 2003 American Chemical Society Published on Web 07/26/2003
Figure 1. Schematic diagram of the experimental setup. The syringe-connected tubes are aligned in the x and y directions, respectively, so that they can manipulate cells toward the Si/SiO2 opening (upper inset), which is surrounded by a suitable microsensor(s). In the Cell Interrogator chip used in this work, a Pt microring electrode was realized, to be used for oxygen monitoring and other amperometric applications. The outer sensor edge and the connecting Al wire are embedded in a continuous polyimide surface that rejects cells as a result of its relative hydrophobicity. The hydrophilic Si/SiO2 opening is exposed from the polyimide coating so that cells can attach there.
(or other suitable microsensors) sitting atop surface-oxidized silicon (Si/SiO2) substrate and encased in micrometer-scale wells (of virtually zero depth) fashioned from a patterned thin polyimide film. At sites within the Pt rings, Si/SiO2 is exposed from the polyimide film. These relatively hydrophilic openings (diameter, d ) 20 µm or 30 µm in this work; that can be easily made smaller if needed for studies of smaller cell clusters or single cells) are used as acute cell attachment sites (Figure 1). Details on microfabrication of such chips are described in ref 12. To test the syringe-and-tube cell manipulation system, a microfabricated chip was placed under PBS solution in a polystyrene Petri dish (Becton Dickinson, 60 × 15 mm). Two sets of syringe (Becton Dickinson, 10 cm3) and Tygon microbore tube (o.d. ) 2.0 mm, i.d. ) 1.4 mm) were used for navigation in directions of the x and y axes, respectively (Figure 1). A syringe and one end of the tube were connected via a needle (Becton Dickinson, 16G 1). The other end of the tube was placed and fixed on the target substrate, pointing toward the cell-trapping site of interest (a Si/SiO2 opening). The target area was located at the intercept of the roughly perpendicular x and y tube axes at ∼5 mm distance from each tube end. We kept this distance to prevent the image around the Si/SiO2 opening from being blurred as a result of optical interference by the tube(s) and also to acquire optimum flow patterns at the region of interest. A reflectance-type microscope (Leitz Secolux, 6-in. × 6-in. inspection microscope) with 10× objective lens (Leitz, NA ) 0.22) was used to observe magnified images around the target attachment site. Images were monitored by a CCD video camera (Bausch & Lomb, TU Camera, JE3010 A) and acquired and stored by a PC with a video capture board. To monitor flow patterns, a blue-colored dye, Evans blue (Allied Chemical) was employed as a flow marker. A video zoom microscope (Edmund Industrial Optics, VZM300 color system)
was used to observe and record flows. Visualized flows were stored as movies (∼5 frames/s) and converted into still images with time labels by the PC. Materials and Cell Culture. All chemicals were of analytical grade from Fisher and Aldrich. Quartz distilled water (18 MΩ cm2) was used to prepare all solutions. Mouse macrophage cells (BAC1.2F5) whose culture and disposal procedures are described in detail in ref 13 were used in the experiments in this work. Procedures. (a) Flow Pattern Tests. Flow patterns in our syringe-and-tube system were explored first to assist in optimizing the design parameters suitable for practical operation. One set of syringe and tube was prepared, and the outlet end of the tube was placed on the bottom of a Petri dish. The patterns were visualized by taking successive images of expansion of Evans blue ejected from the tube at different syringe pushing rates, monitored from both top and side. We note that diffusion during the short time span of these experiments could not possibly contribute to observed dye expansion. (With D ∼ 10-6 cm2/s, diffusional expansion is about (Dt)1/2 ) 10 µm in 1 s). (b) Manipulation Velocity Tests. Correlation between average flow velocity at the exit of the tube and induced cell velocity was also explored. A syringe was pushed at several different constant rates. At each rate, the distance that a settled cell traveled on polyimide during the time of observation was measured under the microscope, and cell travel speed (Vcell) was calculated. The volume pushed out from the tube during this interval was also read from the syringe measure, to calculate the volumetric flow rate (Q) and the average flow velocity (Vav) at the outlet of the tube. (c) Cell Manipulation. Twenty microliters of PBS solution with suspended cells (∼105 cells/mL) was added to the PBS covering the chip, close to the Si/SiO2 opening (trapping site) by a pipettor (Gilson, 5-20µL variable). After waiting for most cells to settle down onto the substrate (10-20 s), which was observed by tuning the focus plane of the microscope to just above the substrate, cell navigation was started. A selected individual target cell was navigated to the Si/SiO2 opening by controlling (pushing or pulling with a proper speed) the x and y syringes. Five minutes later, also by controlling the syringes to include the optimum cell cleaning rate (Vcell ≈ 5-50 µm/s), all other unwanted cells settled on the surrounding polyimide surface were moved away while keeping the target cell (or cells) on the Si/SiO2 opening attached. THEORETICAL CONSIDERATIONS Schematics of flow patterns in the syringe-and-tube system and the adopted coordinate system are shown in Figure 2. Because the distance of the region of interest (ROI) including the Si/SiO2 opening is ∼5 mm from the tube outlet and the size of the ROI is on the order of 800 × 600 µm, the manipulated cells can be considered to be all in the axial vertical (x-z) plane. Thus, it is sufficient to estimate flow patterns in that plane to derive velocities of cells settled on the substrate, meaning that flow velocity at any location of interest can be represented as V(x, z). The manipulated cell velocity (Vcell) is approximated as one-half of the flow at the top of a settled cell, on the basis of the assumption that cells are rolling with neither slipping nor sticking on the chip. To achieve precise cell patterning, ideal cell manipulation speeds should be in the micrometer-per-second range. Such very (13) El-Moatassim, C.; Dubyak, G. R. J. Biol. Chem. 1992, 267, 23664-23673.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4687
Figure 2. Schematics of the side view of hypothesized flow patterns, with the x-z coordinate system defined as follows: The x axis is along the axis of the x tube, x being zero above the center of the Si/SiO2 opening on the chip and pointing away from the tube. The y axis (not shown here but shown in Figure 1) coincides with the axis of the y tube, similar to the definition of the x axis. The manipulated cells are located about 5 mm away from both outlets, the x outlet being, thus, at x ) -5 mm. The z axis is vertically (perpendicular to the substrate) directed upward from the axis of the x tube where z ) 0. The surface of the substrate is thus at z ) -1 mm. V(R, β) represents flow velocity at x ) R mm and z ) β mm. Vav and Vcell are the average flow velocity at the outlet of the tube and the manipulated cell velocity on polyimide, respectively. Cells are hypothesized here as spheres of 10-µm diameter.
Figure 3. Experimental flow patterns typical for syringe-and-tubebased hydrodynamic cell manipulation. Evans blue dye was used as a marker to visualize flow. Three still pictures (snapshots) at different instances were superimposed to make each panel, resulting in earlier flow patterns becoming darker in grayscale. The front contour of each snapshot is emphasized with white lines. Panel A shows typical flow patterns from side view that were found to be similar for the different flow velocities used. Panels B-D show flow patterns from a top view, with fast flow (V(0, 0) ≈ 3 cm/s; panel B), moderate flow (V(0, 0) ≈ 1 cm/s; panel C), and slow flow (V(0, 0) ≈ 0.12 cm/s; panel D). Cross marks (×) represent the approximate position of the Si/SiO2 opening (5 mm distance from the outlet of the x tube.)
slow speeds are realized in our system using basic properties of viscous flow: the macroscopic flow at the height of the outlet of the tube will automatically downgrade to very low speeds in the vicinity of the substrate where settled cells are. This is what makes it possible to use a macroscopic and relatively coarse system realized by pushing and pulling laboratory syringes at millimeterper-second rates, to achieve fine and precise movement of cells on the chip. The use of two orthogonal syringes makes then any cell trajectory feasible with pushing and pulling them in the right order and at the right rate. Flow in every studied case was found to be essentially laminar everywhere in the flow stream. The Reynolds number is already very small inside the tube in practical cell manipulations: Re ∼ 7 for the typical average flow velocity in the tube, Vav ∼ 1 cm/s, corresponding to a volumetric flow rate, Q ∼ 15 µL/s. The flow then expands and changes symmetry upon exit. While axial symmetry is then lost in three dimensions, it is preserved in the horizontal plane. This flow expansion will increase the Reynolds number outside the tube to some extent; the flow, however, is expected to remain laminar because of the shallow solution layer (∼ 3 mm) and low flow rates used. Stationary laminar flow produces parabolic flow patterns in a tube (Figure 2). On the other hand, shear flow above a plane in steady state is linear in the axial vertical plane sufficiently far from the outlet. Flow at the ROI (5 mm from the outlet) must therefore be one between parabolic and linear, that may be approximated as V(0, z) ≈ V(0, 0)(1 - |z|1.5) at z e 0.
side, did not significantly change with increasing flow velocity (Figure 3A), the horizontal patterns obtained from the top view strongly depended on absolute velocity (Figure 3B-D). From a narrow jet (Figure 3B) at relatively high speeds, through moderate broadening (Figure 3C) and fanning (Figure 3D) to almost circular expansion around the tube outlet at very low speeds (not shown), a variety of patterns are feasible, depending on syringe moving rate. This means that a wide range of cell travel velocities can be produced within the same system. Cell Movement Correlated to Syringe Operation. We have recorded movements of cells settled on the polyimide surface at several syringe pushing speeds (Figure 4, data shown as 0). Regression between the average flow velocity at the outlet of the tube and the observed cell velocities resulted in reasonable linearity (Figure 4, r2 ) 0.74). This flow calibration was compared with the cell velocity derived from visually observed flow patterns in Figure 3D. The syringe is operated, in typical cell manipulation, at speeds similar to those corresponding to Figure 3B-D (0.1-3.0 cm/s, slower velocities being used for fine adjustments). However, only images in Figure 3D could be used to estimate V(-5, 0) and V(0, 0), because of limitation in the available temporal resolution at the faster flow rates. The flow patterns yield V(-5, 0) ≈ 0.19 cm/s and V(0, 0) ≈ 0.12 cm/s, resulting in Vav ) (1/2)V(-5, 0) ≈ 0.10 cm/s (Q ) 1.5 µL/s) and Vcell ≈ (1/2)V(0, -0.99) ≈ V(0, 0)(1 |-0.99|1.5) ) 9 µm/s, respectively. (Here, z ) -0.99 mm is used because the approximate cell size is 10 µm.) This result, indicated in Figure 4 as 4, lies well within the 95% confidence interval of the straight line fitted to experimental results from actual cell navigation experiments. This lends support to the simple semiquantitative theoretical considerations outlined above.
RESULTS AND DISCUSSION Vertical and Horizontal Flow Patterns. While the vertical components of representative flow patterns, observed from the 4688 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
Figure 4. Experimental correlation (solid line) between the average flow velocity at the outlet of the tube and the manipulated cell velocity on polyimide observed under the microscope (0), with a result by theoretical derivation from panel D in Figure 3 shown as a triangle (4). Dashed lines represent the boundaries of the 95% confidence interval around the linear regression line.
Selective Cell Navigation and Patterned Attachment on BioMEMS Chips. Figure 5 shows sequential pictures of navigated and then trapped cells, obtained with the hydrodynamic micromanipulation system. Individual cells are identified and tracked by corresponding lower-case letters (a-h). In the sequence, the chosen target cell (a) was navigated in consecutive
discrete steps onto a Si/SiO2 opening of 20-µm diameter by sequentially pushing the x and y syringes. The other cells (b-h) were, of course, also moved parallel to the trajectory of cell a. It required ∼1 min at the most to perform all these navigating steps (Figure 5A-E). After ∼5 min of waiting, those cells that settled on the polyimide surface were removed in the cleaning steps (Figure 5F-G), while the initially chosen target cell stayed trapped on the desired cell site (Figure 5H). It took less than 7 min to accomplish all these procedures. About 5 min of waiting between the navigating steps and the cleaning step(s) was sufficient for keeping the defined target cell(s) attached on the opening while removing all other cells from the surrounding polyimide surface. This observation corroborates our previous results: an incubation period of no longer than 5 min was found to be adequate for developing a sufficient adhesion difference between polyimide and Si/SiO2 surface.12 Figure 5I depicts the entire trajectory of cells a-d, from which it is obvious that the x and y axes of cell travel lines were not exactly perpendicular (inclined at an ∼56° angle). Although this did not prevent the selected cell from reaching the target site, a more orthogonal arrangement of the two tube openings is preferred, since this can ensure optimum efficacy, that is, to achieve the end result in the smallest number of steps. Repeating this single cell navigating, trapping, and cleaning procedure, we also succeeded in making several preselected
Figure 5. Sequential pictures of cell movement on a Cell Interrogator chip using the hydrodynamic manipulation technique. Cells a-h are micromanipulated by the x and y syringes (positioned to the left and the below of this figure, respectively), aiming cell a to be attached onto the Si/SiO2 opening site while rejecting all other cells. The same lower case letters (a-h) label the same cells in each picture. Panel A shows starting conditions after the cells settled on the chip. Each subsequent picture (B-E) shows conditions after one syringe operation step for navigation. After waiting for 5 min, all cells except cell a are moved away (F-H). Panel I shows the resulting trajectories of the individual cells (a-d) during the entire cell manipulation process.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4689
Figure 6. Examples of preselected cell patterrns formed by hydrodynamic cell manipulation. A Si/SiO2 opening site (30 µm diameter) surrounded by a circular Pt ultramicroelectrode is shown with attached cells. A: Y-shape geometry; B: densely packed cells in the opening.
desired formations of small, clearly defined cell clusters. In these experiments, we used a 30-µm-diameter cell attachment site to form the desired monolayer patterns. Two examples are shown in Figure 6 (Figure 6A, Y shape; Figure 6B, densely packed cells). It took about 30 and 50 min, respectively, for the entire procedure to achieve the desired patterns. CONCLUSIONS Several studies regarding multicell or single-cell measurement on micromachined (MEMS) devices with coupled microanalytical sensors have been reported.11,14 The technique described in this work could be a good tool for manipulating single cells to such measuring sites. Having the advantages of easy and gentle cell manipulation and acute (not cell-culture-and-growth-based) pattern formation, the hydrodynamic manipulation technique renders it feasible to realize Cell Interrogator systems where cell sites are surrounded by multiple sensors.9,12 An example is when recording electrochemical variables with MEMS-made microelectrodes or optical signals obtained with transparent glass openings instead of Si/SiO2, aligned with the desired cell architecture. Simplicity of the syringe-and-tube-based micromanipulation technique may also make this approach useful as a tool for gene transfection, drug injection, and other technologies in cellular engineering. It is also straightforward to implement a motorized (14) Bratten, C. D. T.; Cobbold, P. H.; Cooper, J. M. Anal. Chem. 1998, 70, 1164-1170.
4690 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
version of the system, by pushing/pulling syringes with interactive or even fully computer-controlled motors. This can make the procedure even automatic once the desired cell(s) and the target are selected on the computer screen, using then image processing for feedback. Optimum trajectories can then be realized that account for other present cells and even obstructions. Flow speed can also be optimized. Instead of discrete x and y steps, trajectories of any angle can be produced, including curved ones to circumnavigate obstacles. Such a system will be explored in an upcoming publication for cellular studies, as well as for piece-by-piece microand nanofabrication. ACKNOWLEDGMENT The authors acknowledge a Case Prime Fellowship (CWRU) awarded for M.Y. Dr. Christian Zorman’s contribution to the fabrication of the BioMEMS chips, and Dr. George Dubyak’s expertise with cell culture of the macrophage cells are also gratefully acknowledged. Professor Arthur Heuer is thanked for coining the term “Cell Interrogator” to concisely describe the BioMEMS platform reported here.
Received for review February 6, 2003. Accepted June 22, 2003. AC030055U
